YOLOF

YOLOF copied to clipboard

AssertionError: bad box: x1 larger than x2

1

[07/11 18:24:42 d2.engine.train_loop]: Starting training from iteration 0 [07/11 18:24:52 d2.utils.events]: eta: 7:53:07 iter: 19 total_loss: 1.522 loss_cls: 0.2083 loss_box_reg: 0.0001553 loss_mask: 0.6411 loss_rpn_cls: 0.6729 loss_rpn_loc: 0.02641 time: 0.4052 data_time:...

raj-4444

AssertionError: A prediction has class=56, but the dataset only has 4 classes, and the predicted class id should be in [0, 3]. Where in the detectron2 config will I change...

Satyajit1993

AssertionError: Attribute 'thing_dataset_id_to_contiguous_id' in the metadata of 'coco_2017_train' cannot be set to a different value! {1: 0} != {0: 0}

2

AssertionError: Attribute 'thing_dataset_id_to_contiguous_id' in the metadata of 'coco_2017_train' cannot be set to a different value! {1: 0} != {0: 0} 修改detectron2/data/datasets中builtin_meta.py文件，但是训练报错，是不是类别不匹配，configs/default.py文件中的类别数也改过来了。（训练一类数据集）

zkungithub

About datasets

3

我想调试代码，怎么讲GPU的数量和运行网络导入到train文件里面

living1001

KeyError: "No object named 'YOLOF' found in 'META_ARCH' registry!" I hope I can get an answer. Thank you

wzsdnpy

Have anyones trained this code in pascal voc? I had implemented a version of tensorflow codes, but it performed slightly worsh compared with RetinaNet and FCOS. And there is my...

JiXuKong

AssertionError: Attribute 'thing_dataset_id_to_contiguous_id' in the metadata of 'coco_2017_train' cannot be set to a different value! {1: 0} != {0: 0} 修改detectron2/data/datasets中builtin_meta.py文件，但是训练报错，是不是类别不匹配，configs/default.py文件中的类别数也改过来了。（训练一类数据集）

zkungithub

Question on the way implicit objectness is combined with class logits

1

https://github.com/chensnathan/YOLOF/blob/a44b7af0efa9b89d3e91db70bc9a39e49d5d0144/yolof/modeling/decoder.py#L105 Could you please explain why implicit objectness is combined with class logits like this? Thanks.

thisisi3

Awesome work. For datasets with larger scale variations than COCO.Are there any good ways to further optimize YOLOF？ For example, is it feasible to deepen Dilated Encoder? Looking forward to...

QAQEthan

Freeze the Backbone and Dilated Encoder

1

Hi, Many thanks for your awesome work. In the original Detectron2, we can freeze certain layers of the backbone with _C.MODEL.BACKBONE.FREEZE_AT. Can you please let me know how to freeze...

ukdsvl